DeepSeek V4 Pro download MoE architecture 1.6T parameters 1M context window reasoning modes Think Max sparse attention coding agent agentic AI benchmarks API pricing open weights MIT license.
Related
lucidrains/RoboTTT: Implementation of RoboTTT proposed by Yunfan Jiang et al. of Stanford and Nvidia
Implementation of RoboTTT proposed by Yunfan Jiang et al. of Stanford and Nvidia
Harshit-Lalwani/Issue-Assignee-Recommender: Two-stage recommender (BM25/embedding retrieval + LightGBM LambdaRank) that ranks the developers most likely to resolve a newly-filed Jira issue, built and evaluated on 130k real issues from 5 Apache/Jira projects.
Two-stage recommender (BM25/embedding retrieval + LightGBM LambdaRank) that ranks the developers most likely to resolve a newly-filed Jira issue, built and evaluated on 130k real i...